Open AccessComparison of hard limiting and soft computing methods for predicting software effort estimation: In reference to Small Scale Visualization Projects
Author(s) -
T. M. Kiran Kumar,
M. A. Jayaram
Publication year - 2018
Publication title -
international journal of engineering and technology
Language(s) - English
Resource type - Journals
ISSN - 2227-524X
DOI - 10.14419/ijet.v7i4.6.21195
Subject(s) - mean squared error , computer science , ambiguity , software , visualization , soft computing , scale (ratio) , software metric , data mining , estimation , cyclomatic complexity , software development , source lines of code , software quality , machine learning , statistics , mathematics , systems engineering , programming language , engineering , physics , quantum mechanics , artificial neural network
It is a well known fact that software effort estimation is exceptionally critical in every software industry, particular during the development of projects. It is hard to estimate the parameters involved due to ambiguity and uncertainty associated with the parameters. It is exactly here the hard limiting techniques, soft computing techniques comes to play. In this unique circumstance, this paper, presents an attempt to that compare the two paradigms for effort estimation. For this, we have considered fifty real time small visualization projects thrive by post graduate students. The prototype development involves following stages: i) Elicitation of seven novel parameters namely Lines of Code, Cumulative Grade Point Average, New and changed code, Reused code, Cyclomatic Complexity, Algorithmic Complexity and Functional Points. ii) Developing of hard limiting methods and soft computing methods for prediction of software effort involved in terms of duration in minutes.For the validation of the models error metrics namely: Mean Absolute Error (MAE), Mean Magnitude of Relative Error (MMRE), Mean of Magnitude of error Relative to the Estimate (MMER) and Root Mean Square Error (RMSE) have been used. The result showed that the models compared very well with marginal difference in terms of predict values of error matrix.